BacA is responsible for phosphorylation of undecaprenyl in E. coli and other bacteria. This is an important step in peptidoglycan synthesis. Studies on homologues of E. coli bacA in S. aureus and S. pneumoniae revealed that gene deletion mutants were attenuated in mouse models of infection [39]. We identified homologues in the genomes of M. smegmatis and M. tuberculosis (Upk; Fig. 1), and investigated the role of the cognate Upk in mycobacteriae, in vitro and in vivo. In addition, downstream consequences of the gene deletion in M. tuberculosis were investigated globally on the transcriptome and proteome level.

Impact of upk deletion on cell wall attributes

S. aureus and S. pneumoniaeΔbacA mutants show no significant alterations in growth rate or morphology [39]. In contrast, the in-frame, unmarked deletion of upk (Fig. 3), in M. smegmatis revealed a distinct phenotype (Fig. 4). While the growth rate was almost unaltered, M. smegmatis Δupk colonies did not display dome-like structures, like that of the wildtype strain. Rather, they showed caved-in structures indicating autolysis of the cells. Upk deletion must be responsible for the altered colony morphology since complementation resulted in a return to dome-like colonies. A comparable phenomenon did not appear for M. tuberculosis or M. bovis BCG. This suggests that Upk may play different roles, or at least have different physiological functions, in different mycobacteria.

An altered cell wall might exhibit visible changes when examined by electron microscopy, however, the M. smegmatisΔupk cell wall appeared comparable to wildtype by electron microscopic examination (Fig. 8). More direct analyses of the cell wall by immuno-gold staining revealed less, though still detectable, peptidoglycan in the mutant. Deletion of the upk gene, therefore, did not lead to complete loss of peptidoglycan in the cell wall. Although the effect observed was small, alternative, yet less efficient pathways for undecaprenyl-phosphorylation or transport of peptidoglycan precursors may exist, as indicated by the slightly slower growth of M. smegmatisΔupk mutant in vitro (Fig. 9).

As described for E. coli, S. aureus,and S. pneumoniae, undecaprenyl phosphokinase deletion mutants exhibit higher sensitivity to bacitracin [39,38]. A knockout mutant strain of the mycobacterial homologue upk was expected to show the same susceptibility. This was verified unequivocally by the alamar blue assay for M. smegmatis (Fig. 10). Complementation with the M. tuberculosis upk homologue rv2136c partially reversed the susceptibility of M. smegmatis. Incomplete complementation may be due to constitutive gene expression of upk driven by the groEL2 (Hsp60) promoter, which does not reflect the physiological regulation of the gene. Surprisingly, the M. tuberculosis mutant strain was the first described bacterial upk deletion mutant that did not exhibit altered sensitivity to bacitracin. This finding demonstrates the uniqueness of the M. tuberculosis cell envelope not only to other bacteria but also to fast growing non-pathogenic species of mycobacteria [24,69], and emphasizes that even highly conserved proteins can have a range of activities in different species. M. tuberculosis may possess an efficient alternative pathway to shuttle out peptidoglycan precursors or may preserve the integrity of its cell wall by overproduction of other components. However, growth properties of M. tuberculosisΔupk in pellicle cultures were affected (Fig. 18) and this is likely due to altered surface features.

The reconstituted strains in this study were constructed on the knockout background by electroporation with the episomal multi copy plasmid pMV262-upk. In all cases, the M. tuberculosis H37Rv upk gene was under the control of the M. bovis groEL2 (Hsp60) promoter. With regard to distinct properties, the reconstituted strain failed to display unilaterally a wildtype phenotype: M. smegmatisΔupk + pMV262-upk grew in dome-like colonies and persisted in macrophages similar to wildtype M. smegmatis, whereas resistance to bacitracin was intermediate between wildtype and Δupk mutant.

M. tuberculosis mouse infections never displayed a wildtype-like phenotype for M. tuberculosis Δupk + pMV262-upk. This phenomenon is likely due to weak transcriptional activity. For various genes the groEL2 promoter can provide the cell with more transcripts than the natural promoter, but for late log phase in vitro the groEL2 promoter transcribed about 200 fold weaker than the M. tuberculosis wildtype promoter. This severe regulation problem provides the most likely explanation for the failure of the complementation strain to achieve a wildtype phenotype in infection. In the case of M. tuberculosis pellicle formation, an intermediate phenotype of the complementation strain compared to wildtype and Δupk mutant was also observed.

Physiological balance

Upk is thought to be critical for mycobacterial cell envelope formation and it was, therefore, expected that the Δupk mutant would attempt to compensate in some way. Global screening by proteome and transcriptome analysis was performed to identify putative compensatory mechanisms. One obvious response of the M. tuberculosisΔupk mutant was the upregulation of a FASII-system related operon (rv2243 – rv2247; Fig. 21). The involvement of KasA / Rv2245, one gene product of this operon, in Isoniazid resistance, has been thoroughly investigated. A report by Mdluli et al. on Isoniazid-resistant patient isolates, which lacked other mutations associated with resistance to the drug, showed amino acid altering mutations in the KasA protein [70]. Additional studies further supported a role of KasA and KasB in Isoniazid resistance [67]. In contrast, other studies [71,72] reported that 3 of the 4 clinical isolates bearing mutated kasA-alleles were fully susceptible to Isoniazid. To date, gene transfer experiments determining whether any of these mutations confers Isoniazid resistance to susceptible strains of mycobacteria have not been performed. We show that the Δupk deletion mutant of M. tuberculosis H37Rv, which overexpressed kasA, did not exhibit increased resistance to Isoniazid in the alamar blue assay (Fig. 22B), but rather had a slightly higher susceptebility. This finding supports a recent study which favors a gene distinct from kasA, namely inhA, as the major primary target for Isoniazid [66].

Mycobacterial FAS-II, unlike other bacterial type II FAS cognates, is incapable of de-novo fatty acid biosynthesis [73], however it appears able to elongate C₁₄-AcpM of mycobacteria and C₁₆-AcpM to preferentially long chain fatty acids ranging from 24 to 56 carbon atoms. Recent studies suggest that KasA (Rv2245) is part of FAS-II and participates in mycolic acid biosynthesis [74,75]. Mycolic acids are high molecular weight α-alkyl, β-hydroxy fatty acids with the general structure R-CH(OH)-CH(R’)-COOH, where R is a meromycolate chain consisting of 50 – 56 carbons and R’ is a shorter aliphatic branch possessing 22 – 26 carbons [24]. Mycolic acids are key components of the mycobacterial cell wall (Fig. 2) and play a role in producing an effective lipophilic barrier. Considering the importance of mycolic acids in bacterial survival and maintenance of cell wall integrity, the Δupk deletion mutant may benefit from overproduction of this cell wall component in order to overcome reduced peptidoglycan. In addition, increased mycolic acids may improve the blocking of phagosome-lysosome fusion and better allow the bacteria to escape degradation by the host. This strategy prevents exposure of the bacterium to the hostile environment of the lysosome while rendering it accesible to nutrients endocytosed by the cell. The underlying mechanism is not yet fully understood [76,77]. This function has been proposed and demonstrated for other cell envelope components, such as trehalose 6,6'-Dimycolate (TDM), a mycobacterial glycolipid cord factor. TDM has also been implicated in interfering with phagosome-lysosome fusion [78].

Infection of macrophages with M. smegmatis can be used as a model to analyze bacterial persistence in the host [30]. Lower persistence of the M. smegmatisΔupk mutant in macrophages indicates a role of Upk in mycobacterial virulence/persistence. Indeed, the M. smegmatisΔupk mutant was cleared more rapidly from the host cells. These findings indicate the importance of a robust cell envelope for persistence in the host. In this experiment complementation with the M. tuberculosis upk gene was sufficient to revert from mutant to wildtype phenotype. However, it is not valid to generalize knowledge gained from experiments with M. smegmatis. In certain aspects, M. tuberculosis is unique as was demonstrated in the case of resistance to bacitracin.

Biofilm formation

The term biofilm describes a population or community of bacteria living in organized structures at a liquid interface. Early confocal laser scanning microscopy (CLSM) of single species biofilms [79,80] revealed that biofilm bacteria live in cellular clusters or microcolonies that are encapsulated in a matrix composed of an extracellular polymeric substance (EPS), separated by open water channels which act as primitive circulatory system for the delivery of nutrients and removal of metabolic waste products. Within a biofilm, each bacterium occupies a specific microenvironment, which is defined by the surrounding cells, the proximity to a channel and the EPS matrix. The structuring of biofilms in microcolonies and fluid channels has been shown to be influenced by fluid flow, nutrient composition, and intracellular small messenger molecules, which are used for bacterial communication [81,82,83].

Various gram-negative and gram-positive bacteria, as well as fungi, grow in two forms: planktonic and, as a step of microbial development, in a biofilm [33]. M. smegmatis and other non-tuberculous mycobacteria such as Mycobacterium fortuitum and Mycobacterium marinum live and grow planktonicly or as a biofilm [34]. Biofilms support resistance to antimicrobial chemotherapy and play a role in contamination in clinical and industrial settings. Biofilm formation poses a major problem because they can increase drug resistance [36]. The reduced metabolic and growth rates shown by biofilm bacteria, particularly those deep within the biofilm, can render these microbes inherently less susceptible to antibiotics. The EPS matrix can act as an absorbent or reactant, thereby reducing the amount of drug available for action on biofilm cells. Moreover, biofilm bacteria are physiologically distinct from their planktonic cognates and express specific protective factors such as multidrug efflux pumps and stress response regulons [84,85].

Biofilm growth of M. smegmatis was unaffected at Isoniazid concentrations that inhibited growth of planktonic bacilli [35]. Previously described deletion mutants of M. smegmatis lacking the capability of glycopeptidolipid acetylation, which affects the cell envelope, are defective in biofilm formation [86,60]. The in-frame, unmarked deletion mutant of the M. smegmatis upk gene is the first evidence for a role of Upk in biofilm formation. Upon adherence, the Δupk mutant strain formed a scattered biofilm only. Adherence could have been reduced due to a missing extracellular matrix (Fig. 13) and to slightly inferior growth-properties in biofilm medium.

In vivo

The apathogenic environmental M. smegmatis owes its name from isolation from genital secretions (smegma): In November 1884, Lustgarten reported to the Royal Society of Medicine in Vienna that he had discovered a bacterium with staining characteristics of tubercle bacilli in syphilitic chancres and gummae [31]. Soon thereafter Alvarez and Tavel identified microorganisms similar to those in normal genital secretions (smegma) [32]. Smegma is associated with hygienic conditions and has been proposed as risk factor for penile cancer [87,88]. To determine whether Upk plays a role in genital smegma development by M. smegmatis we developed an in vivo model for M. smegmatis biofilm formation. In this in vivo mouse model of M. smegmatis biofilm formation the Δupk deletion mutant was found to be deficient in induction of smegma development (Fig. 14), thus stressing the relevance of Upk in mycobacterial saprophytic life.

The giv phenotype

A direct consequence of the upk gene deletion in M. tuberculosis is the expression of a growth in vivo (giv) mutant phenotype. Studies on defined mutants of M. tuberculosis in the mouse model of infection have led to the classification of attenuated mutants in several phenotypic classes [89]. These mutants have been categorized by their growth characteristics, namely: i) severe growth in vivo (sgiv) mutants, which show a marked reduction in colony-forming units over time; ii) growth in vivo (giv) mutants, which grow less robustly than wildtype M. tuberculosis in the lungs of immunocompetent mice, yet still grow better than sgiv mutants; iii) persistence (per) mutants, which fail to grow or persist after the onset of acquired immunity, and iv) mutants with the same growth characteristics as per mutants, but show altered pathology (pat) compared with that of wildtype M. tuberculosis. Most mutants, including the Δupk mutant, fall into the giv class, showing reduced growth (Fig. 23) and pathology (Fig. 25), resulting in an attenuated phenotype and increased survival of infected immunocompromised mice (Fig. 27). Examples for giv mutants of M. tuberculosis are the two component regulatory protein phoP [90], the accessory secretion factor secA2 [91], the glutamine synthase glnA1 [92], and panCD [93], which is involved in pantothenate synthesis. The gene encoding the exported repetitive protein (erp) in both M. bovis BCG and M. tuberculosis has no ascribed function and is specific for mycobacterial species [94]. Deletion of the erp gene results in impaired growth of bacilli in lungs and spleens of infected mice, as with the Δupk mutant. Berthet et al. postulated that virulence depends on the ability of the bacilli to multiply [95]. The virulence reduced M. tuberculosisΔupk mutant exhibited characteristics of a giv mutant, which could render it an interesting vaccine candidate, because it maintains replication and is more likely to result in a long-lasting host immune response compared to a sgiv mutant. However, a potentially dangerous situation could arise if the mutant regains its rate of growth, as may happen in immunocompromised individuals (Fig. 27), resulting in disease. Thus, continued characterization of specific mutants of M. tuberculosis is required to develop strains which elicit a strong protective immune response, but fail to reactivate in immunodeficiet individuals.

The giv phenotype of the Δupk mutant, which was related to reduced growth in vivo, is probably related to its altered cell envelope. Forty percent of the most significantly upregulated genes in the Δupk mutant compared to wildtype were related to cell envelope processes and belonged to the tuberculist categories “lipid metabolism”, and “cell wall and cell processes” (Fig. 20). Hence, impaired self protection of the tubercle bacilli, as consequence of the impaired cell wall, or improved accessibility of antigens, or upregulation of one or more antigens which resulted in better processing and recognition by T cells, and therefore a more potent immune response, could allow improved control by the host. Furthermore, the M. tuberculosisΔupk mutant may switch prematurely to a dormancy program as may indicated by abundance of the HspX protein, a marker of M. tuberculosis latency [96,64].

In addition, host effector mechanisms may be more effective against the altered cell wall of M. tuberculosisΔupk than against M. tuberculosis wildtype. Such mechanisms may include production of highly reactive low molecular weight molecules, in particular reactive oxygen intermediates (ROI) and reactive nitrogen intermediates (RNI) [97,98] (Fig. 1). ROI and RNI cause damage of cellular constituents by oxidation of cellular membranes and enzymes, DNA damage, mutagenesis, and inhibition of membrane transport processes [99,100,18].

The impaired cell wall structure of the upk deficient strain, could form a weaker barrier against these effector molecules with the possible consequence of enhanced killing of the pathogen by the host. A larger quantity of antigen that could be processed or a higher accessibility of antigens may further contribute to increased susceptibility of M. tuberculosisΔupk to the adaptive immune response.

Vaccine

As discussed above, the giv mutant phenotype of M. tuberculosisΔupk represents a promising phenotype for development of an attenuated M. tuberculosis mutant that could serve as a potential vaccine candidate. Nevertheless, a single gene deletion mutant in M. tuberculosis is unlikely to fulfill the safety standards required for a vaccine to be used in humans notably in immunocompromised individuals. Thus, a giv mutant like M. tuberculosisΔupk, needs to be thoroughly characterized and subsequently refined in further steps of development before it could become a valid vaccine candidate.

The vaccine strain M. bovis BCG offers an impressive safety record but unsatisfactory protection [101]. In this study, we deleted the upk gene from M. bovis BCG. This is an alternative strategy that relies on the basic premise that M. bovis BCG could be re-engineered to enhance its efficacy. For example, recombinant M. bovis BCG expressing various cytokines have been shown to improve the response against M. tuberculosis antigens [102] and recombinant M. bovis BCG expressing listeriolysin of Listeria monocytogenes showed an enhanced capacity to stimulate CD8⁺ T cells [103]. In addition, a recombinant M. bovis BCG strain that overexpresses the 30 kDa Ag85 protein has been reported to provide an improved protection against M. tuberculosis infection [104].

The upk deletion which resulted in an even more attenuated strain, was not expected to cause safety problems and accordingly used in a vaccine trial. Considering the lower bacterial load of M. bovis BCG Δupk upon vaccination (Fig. 28) and the delayed induction of the IFNγ response (Fig. 29) it was surprising that the modified vaccine strain was able to induce a significantly improved long-lasting protection against M. tuberculosis infection (Fig. 30). This phenotype is difficult to explain at present. Perhaps immunorelevant antigens were overexpressed to balance upk deficiency, or improved killing of M. bovis BCG Δupk provides the immune system with a larger amount of antigens to be processed, resulting in an improved adaptive immune response.

Outlook

The mouse model of smegma was newly established and offers the opportunity for further development. The structure of smegmata can be investigated by microscopy and therefor allows the following questions to be addressed:

Is it possible to stain bacteria within the smegma?

Are the bacteria organized within the smegma?

As for biofilms, are channels existing for facilitated supply with nutrients?

Furthermore, the background growth which arose upon treatment, should be characterized to determine whether it contains other bacteria which induced genital secretions.

The reconstituted knockout strain of M. tuberculosis failed to exhibit the characteristics of a wildtype strain. Polar effects of the knockout construct may have disturbed transcription of downstream genes. It is also possible that the observed M. tuberculosisΔupk phenotype is due to disruption of the expression of an operon rather than a single gene. Transcription of putatively affected genes should be investigated by RT-PCR. And in the case of a disrupted operon, a reconstituted strain that complements all affected genes should be constructed.

It is still an unsolved problem why M. tuberculosisΔupk remains resistant to bacitracin. Perhaps the antibiotic does not have access to its target in M. tuberculosis. Further tests to determine resistance/susceptibility to antibiotics other than bacitracin should be performed, as this may be an important question to consider for drug development against M. tuberculosis. Alternative functions of Upk should also be considered in regards to the mode of action of differentially active antibiotics.

The results derived from proteomic and transcriptomic analyses of M. tuberculosisΔupk strongly suggest an altered cell wall composition. An obvious experiment to follow this would be to extract lipids and to compare the differences between wildtype and mutant M. tuberculosis.

The promising results of the vaccine trial with M. bovis BCG Δupk advocate further development of the M. tuberculosisΔupk mutant strain. A M. tuberculosis mutant might induce a better protection against M. tuberculosis wildtype infections because it would prime the immune system with the same antigens as the causative agent M. tuberculosis, including antigens that are missing in M. bovis BCG like the 129 antigens of the 16 regions of deletion which were lost when Calmette and Guérin generated the vaccine strain from M. bovis. Nevertheless, for safety reasons, a one gene deletion mutant of M. tuberculosis will never be considered as vaccine candidate. Another phase of development is needed to further attenuate the strain and keep its protective potential at the same time. Nevertheless, the promising results of the vaccine trial with M. bovis BCG Δupk suggest this strain has potential for future development as a vaccine candidate. Survival experiments with immunocompromised animals would be necessary to demonstrate its safety. In order to further investigate whether optimal protection is due to persistence of the vaccine strain, regulatory mechanisms of the immune system, or the timing/duration of IFNγ response upon vaccination, additional modified vaccine trials need to be performed.